Artificial intelligence is being introduced into all areas of business and is changing the way innovative companies work. Our team of Machine Learning and AI experts is dedicated to understanding your unique business needs and crafting custom AI and Machine Learning solutions that fit seamlessly into your workflows or implement commercially available AI tools like ChatGPT, IBM Watsonx.aI or countless others tailored to your business. Whether you are a startup, SMB, or enterprise, 7T AI Catalyst scales to meet your demands.
AI and Machine Learning
Visa Application Processing
Visa Application Processing via Multimodal Learning
Our team has developed a solution to process various types of visa applications, from I29 to others, by leveraging multimodal learning. This system handles the entire visa application process, including extraction and analysis of information from passports and resumes.
One of the significant challenges we faced was the dynamic nature of visa application forms, with questions being added and removed constantly. Our solution was a dynamic deep-learning model capable of adapting to these changes and providing accurate results regardless of the form modifications. Moreover, we also devised a system to extract information from more than 200+ different passports, a testament to our team’s adaptability and innovative thinking.
Invoice Automation
End-to-End Invoice Automation via Multimodal Learning
Our team has successfully developed an all-encompassing solution for invoice automation powered by multimodal learning. This project involved designing and constructing a sophisticated deep learning and machine learning pipeline capable of processing logistics invoices from initiation to conclusion. This pipeline is supported by six bespoke deep learning models, each specifically designed to address various aspects of invoice processing.
A critical challenge our team successfully overcame involved the design and development of a custom machine learning operations (MLOps) system and an automatic retraining mechanism. This enabled our system to adapt and learn from new data patterns, making it a truly self-evolving solution. We have managed to handle an impressive volume of 50,000 documents every 12 hours, thanks to our optimal use of PyTorch, Python, and C++.
Invoice Automation
End-to-End Invoice Automation via Multimodal Learning
Our team has successfully developed an all-encompassing solution for invoice automation powered by multimodal learning. This project involved designing and constructing a sophisticated deep learning and machine learning pipeline capable of processing logistics invoices from initiation to conclusion. This pipeline is supported by six bespoke deep learning models, each specifically designed to address various aspects of invoice processing.
A critical challenge our team successfully overcame involved the design and development of a custom machine learning operations (MLOps) system and an automatic retraining mechanism. This enabled our system to adapt and learn from new data patterns, making it a truly self-evolving solution. We have managed to handle an impressive volume of 50,000 documents every 12 hours, thanks to our optimal use of PyTorch, Python, and C++.
CAD Extraction
Structured Information Extraction from CAD Documents
We’ve developed a solution capable of extracting structured information from complex and large CAD documents. Given the substantial size and detailed nature of these files, this was a particularly challenging task. Despite these challenges, our system processes an astounding 15,000 CAD documents daily, meticulously extracting all entity and layout information.
The main challenge was dealing with the substantial file size and complexity of CAD documents. We successfully tackled this issue by developing efficient algorithms to parse and extract information without compromising the integrity of the original data.
Semantic Search
Contextual Semantic Search Engine for Documents
Our team has engineered a semantic search engine that not only locates the queried text within documents but also provides the specific page number and paragraph. This project involved creating an intelligent system capable of understanding the context and semantics behind queries, enabling it to deliver precise results.
Currently, we are expanding this technology into a chat mode that will respond to queries similarly to the ChatGPT model developed by OpenAI, complete with citation references.
The primary challenge in this project was the development of an algorithm capable of understanding and responding to semantic and contextual nuances in user queries. Our solution was to leverage advanced natural language processing techniques to decode these nuances and provide accurate results.
Semantic Search
Contextual Semantic Search Engine for Documents
Our team has engineered a semantic search engine that not only locates the queried text within documents but also provides the specific page number and paragraph. This project involved creating an intelligent system capable of understanding the context and semantics behind queries, enabling it to deliver precise results.
Currently, we are expanding this technology into a chat mode that will respond to queries similarly to the ChatGPT model developed by OpenAI, complete with citation references.
The primary challenge in this project was the development of an algorithm capable of understanding and responding to semantic and contextual nuances in user queries. Our solution was to leverage advanced natural language processing techniques to decode these nuances and provide accurate results.
Machine Learning Powered AR Image Processing for Accurate Body Measurements
AR Image Processing for Accurate Body Measurement
7T Created an Augmented Reality (AR) to simplify the online apparel shopping experience. It uses the consumer’s smartphone camera to provide quick, simple body measurements, creating a 3D “fit model” that recommends the shopper’s size across the platform’s available retailers or dimensions to build custom clothing.
To create a training dataset, we gathered 200+ properly posed scans using a phone camera in a controlled environment. We also took body measurements of the same people with a measuring tape. The scans were then processed through a set of algorithms which involved body segmentation using DeepLab TensorFlow model, body joint location detection using PoseNet model, and various other OpenCV image processing tools and Haar Cascades.
The output of the above process is a set of pixel measurements for various joint to joint lengths. We used the person’s measured height as a calibration to convert pixels to inches. We mapped each tape measurement to a combination of different joint measurements to find the correct weights using Linear Regression that minimizes the error for the sample dataset. This process was repeated and tested with multiple datasets to find a weight for each measurement that would work for any body size and shape with minimal error.
let's begin your digital transformation
7T (also known as SevenTablets) is a Dallas Digital Transformation company, specializing in AI, machine learning and the development of cutting-edge business solutions. Beyond multimodal machine learning and artificial intelligence (AI) development, we also specialize in enterprise software development, mobile app development, web app platforms, process automation, cloud solutions and DevOps, amongst others. We go beyond the role of a typical third-party service provider, taking on the role of a true business partner. 7T takes a "business first" approach to Digital Transformation, with business strategy guiding the development process. ROI is our currency as we solve hard problems with innovative Digital Transformations like machine learning-driven AI development solutions that propel clients to the next level.
